[bug] Simplified lesson (AI) produces a nonsensical lesson

In theory the feature of a simplified lesson is a powerful and useful concept. I was hoping this would help me in understanding the more lengthy podcasts that I use to accumulate more vocabulary. For one lesson it did match as far as I could tell. That podcast was not too long and the summary of the simplified lesson seems ok.

A few days ago I did the simplified lesson for a Finnish description of prehistorical development in the area of Finland and it was just a mess. The first paragraph was taken out of a language learning podcast or article. Then were 2 paragraphs that might have come from the original post (the podcast). Subsequently was a paragraph apparantly written by a commentator on the podcast with praise for such interesting historical material. After that some comment of the podcaster, followed by what seems to be resuming historical matters. What is going on here?

Had lingq fed the original text to the AI, it is improbable that it would have produced this garbage. It seems like the AI was fed the original podcast (which has only audio) where comments could have caused the 4th paragraph praising the podcaster. The first paragraph may have been a random text of some other youtube video (which youtube always shows on the side). I don’t know what else to say. I hope for a speedy repair.

The lesson this happened in is a private lesson “Login - LingQ” imported from a podcast of the channel “TheyTalkSuomi”.

Can you please repair this feature and make sure it is based only on the original text in LIngq?

1 Like

Thanks, we will look into this.

1 Like